1 | ###########################################################################
|
---|
2 | #
|
---|
3 | # LOMPlugin.pm -- plugin for import the collection from LOM
|
---|
4 | #
|
---|
5 | # A component of the Greenstone digital library software
|
---|
6 | # from the New Zealand Digital Library Project at the
|
---|
7 | # University of Waikato, New Zealand.
|
---|
8 | #
|
---|
9 | # Copyright (C) 2005 New Zealand Digital Library Project
|
---|
10 | #
|
---|
11 | # This program is free software; you can redistribute it and/or modify
|
---|
12 | # it under the terms of the GNU General Public License as published by
|
---|
13 | # the Free Software Foundation; either version 2 of the License, or
|
---|
14 | # (at your option) any later version.
|
---|
15 | #
|
---|
16 | # This program is distributed in the hope that it will be useful,
|
---|
17 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
18 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
19 | # GNU General Public License for more details.
|
---|
20 | #
|
---|
21 | # You should have received a copy of the GNU General Public License
|
---|
22 | # along with this program; if not, write to the Free Software
|
---|
23 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
24 | #
|
---|
25 | ###########################################################################
|
---|
26 |
|
---|
27 | ### Note this plugin currently can't download source documents from outside if you are behind a firewall.
|
---|
28 | # Unless, you set the http_proxy environment variable to be your proxy server,
|
---|
29 | # and set proxy_user and proxy_password in .wgetrc file in home directory.
|
---|
30 | # (does that work on windows??)
|
---|
31 |
|
---|
32 | package LOMPlugin;
|
---|
33 |
|
---|
34 | use extrametautil;
|
---|
35 | use ReadTextFile;
|
---|
36 | use MetadataPass;
|
---|
37 | use MetadataRead;
|
---|
38 | use util;
|
---|
39 | use FileUtils;
|
---|
40 | use XMLParser;
|
---|
41 | use Cwd;
|
---|
42 |
|
---|
43 | # methods with identical signatures take precedence in the order given in the ISA list.
|
---|
44 | sub BEGIN {
|
---|
45 | @ISA = ('MetadataRead', 'ReadTextFile', 'MetadataPass');
|
---|
46 | }
|
---|
47 |
|
---|
48 | use strict; # every perl program should have this!
|
---|
49 | no strict 'refs'; # make an exception so we can use variables as filehandles
|
---|
50 |
|
---|
51 |
|
---|
52 | my $arguments =
|
---|
53 | [ { 'name' => "process_exp",
|
---|
54 | 'desc' => "{BasePlugin.process_exp}",
|
---|
55 | 'type' => "string",
|
---|
56 | 'deft' => &get_default_process_exp(),
|
---|
57 | 'reqd' => "no" },
|
---|
58 | { 'name' => "root_tag",
|
---|
59 | 'desc' => "{LOMPlugin.root_tag}",
|
---|
60 | 'type' => "regexp",
|
---|
61 | 'deft' => q/^(?i)lom$/,
|
---|
62 | 'reqd' => "no" },
|
---|
63 | { 'name' => "check_timestamp",
|
---|
64 | 'desc' => "{LOMPlugin.check_timestamp}",
|
---|
65 | 'type' => "flag" },
|
---|
66 | { 'name' => "download_srcdocs",
|
---|
67 | 'desc' => "{LOMPlugin.download_srcdocs}",
|
---|
68 | 'type' => "regexp",
|
---|
69 | 'deft' => "",
|
---|
70 | 'reqd' => "no" }];
|
---|
71 |
|
---|
72 | my $options = { 'name' => "LOMPlugin",
|
---|
73 | 'desc' => "{LOMPlugin.desc}",
|
---|
74 | 'abstract' => "no",
|
---|
75 | 'inherits' => "yes",
|
---|
76 | 'args' => $arguments };
|
---|
77 |
|
---|
78 |
|
---|
79 |
|
---|
80 | my ($self);
|
---|
81 | sub new {
|
---|
82 | my $class = shift (@_);
|
---|
83 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
---|
84 | push(@$pluginlist, $class);
|
---|
85 |
|
---|
86 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
---|
87 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
---|
88 |
|
---|
89 | $self = new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists);
|
---|
90 |
|
---|
91 | if ($self->{'info_only'}) {
|
---|
92 | # don't worry about creating the XML parser as all we want is the
|
---|
93 | # list of plugin options
|
---|
94 | return bless $self, $class;
|
---|
95 | }
|
---|
96 |
|
---|
97 | #create XML::Parser object for parsing dublin_core.xml files
|
---|
98 | my $parser = new XML::Parser('Style' => 'Stream',
|
---|
99 | 'Handlers' => {'Char' => \&Char,
|
---|
100 | 'Doctype' => \&Doctype
|
---|
101 | });
|
---|
102 | $self->{'parser'} = $parser;
|
---|
103 |
|
---|
104 | $self->{'extra_blocks'} = {};
|
---|
105 |
|
---|
106 | return bless $self, $class;
|
---|
107 | }
|
---|
108 |
|
---|
109 | sub get_default_process_exp {
|
---|
110 | my $self = shift (@_);
|
---|
111 |
|
---|
112 | return q^(?i)\.xml$^;
|
---|
113 | }
|
---|
114 |
|
---|
115 |
|
---|
116 | sub can_process_this_file {
|
---|
117 | my $self = shift(@_);
|
---|
118 | my ($filename) = @_;
|
---|
119 |
|
---|
120 | if ($self->SUPER::can_process_this_file($filename) && $self->check_doctype($filename)) {
|
---|
121 | return 1; # its a file for us
|
---|
122 | }
|
---|
123 | return 0;
|
---|
124 | }
|
---|
125 |
|
---|
126 | sub metadata_read {
|
---|
127 | my $self = shift (@_);
|
---|
128 | my ($pluginfo, $base_dir, $file, $block_hash,
|
---|
129 | $extrametakeys, $extrametadata, $extrametafile,
|
---|
130 | $processor, $gli, $aux) = @_;
|
---|
131 |
|
---|
132 | my $outhandle = $self->{'outhandle'};
|
---|
133 |
|
---|
134 | # can we process this file??
|
---|
135 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
---|
136 | return undef unless $self->can_process_this_file_for_metadata($filename_full_path);
|
---|
137 |
|
---|
138 | $file =~ s/^[\/\\]+//; # $file often begins with / so we'll tidy it up
|
---|
139 |
|
---|
140 | print $outhandle "LOMPlugin: extracting metadata from $file\n"
|
---|
141 | if $self->{'verbosity'} > 1;
|
---|
142 |
|
---|
143 | my ($dir,$tail) = $filename_full_path =~ /^(.*?)([^\/\\]*)$/;
|
---|
144 | $self->{'output_dir'} = $dir;
|
---|
145 |
|
---|
146 | eval {
|
---|
147 | $self->{'parser'}->parsefile($filename_full_path);
|
---|
148 | };
|
---|
149 |
|
---|
150 | if ($@) {
|
---|
151 | print $outhandle "LOMPlugin: skipping $filename_full_path as not conformant to LOM syntax\n" if ($self->{'verbosity'} > 1);
|
---|
152 | print $outhandle "\n Perl Error:\n $@\n" if ($self->{'verbosity'}>2);
|
---|
153 | return 0;
|
---|
154 | }
|
---|
155 |
|
---|
156 | $self->{'output_dir'} = undef;
|
---|
157 |
|
---|
158 | my $file_re;
|
---|
159 | my $lom_srcdoc = $self->{'lom_srcdoc'};
|
---|
160 |
|
---|
161 | if (defined $lom_srcdoc) {
|
---|
162 | my $dirsep = &util::get_re_dirsep();
|
---|
163 | $lom_srcdoc =~ s/^$base_dir($dirsep)//;
|
---|
164 | $self->{'extra_blocks'}->{$file}++;
|
---|
165 | $file_re = $lom_srcdoc;
|
---|
166 | }
|
---|
167 | else {
|
---|
168 | $file_re = $tail;
|
---|
169 | }
|
---|
170 |
|
---|
171 | # Indexing into the extrameta data structures requires the filename's style of slashes to be in URL format
|
---|
172 | # Then need to convert the filename to a regex, no longer to protect windows directory chars \, but for
|
---|
173 | # protecting special characters like brackets in the filepath such as "C:\Program Files (x86)\Greenstone".
|
---|
174 | $file_re = &util::filepath_to_url_format($file_re);
|
---|
175 | $file_re = &util::filename_to_regex($file_re);
|
---|
176 | $self->{'lom_srcdoc'} = undef; # reset for next file to be processed
|
---|
177 |
|
---|
178 | &extrametautil::addmetakey($extrametakeys, $file_re);
|
---|
179 | &extrametautil::setmetadata($extrametadata, $file_re, $self->{'saved_metadata'});
|
---|
180 | if (defined $lom_srcdoc) {
|
---|
181 | # copied from oaiplugin
|
---|
182 | if (!defined &extrametautil::getmetafile($extrametafile, $file_re)) {
|
---|
183 | &extrametautil::setmetafile($extrametafile, $file_re, {});
|
---|
184 | }
|
---|
185 | #maps the file to full path
|
---|
186 | &extrametautil::setmetafile_for_named_file($extrametafile, $file_re, $file, $filename_full_path);
|
---|
187 | }
|
---|
188 |
|
---|
189 | return 1;
|
---|
190 | }
|
---|
191 |
|
---|
192 | sub check_doctype {
|
---|
193 | $self = shift (@_);
|
---|
194 |
|
---|
195 | my ($filename) = @_;
|
---|
196 |
|
---|
197 | if (open(XMLIN,"<$filename")) {
|
---|
198 | my $doctype = $self->{'root_tag'};
|
---|
199 | ## check whether the doctype has the same name as the root element tag
|
---|
200 | while (defined (my $line = <XMLIN>)) {
|
---|
201 | ## find the root element
|
---|
202 | if ($line =~ /<([\w\d:]+)[\s>]/){
|
---|
203 | my $root = $1;
|
---|
204 | if ($root !~ $doctype){
|
---|
205 | close(XMLIN);
|
---|
206 | return 0;
|
---|
207 | }
|
---|
208 | else {
|
---|
209 | close(XMLIN);
|
---|
210 | return 1;
|
---|
211 | }
|
---|
212 | }
|
---|
213 | }
|
---|
214 | close(XMLIN);
|
---|
215 | }
|
---|
216 |
|
---|
217 | return undef; # haven't found a valid line
|
---|
218 |
|
---|
219 | }
|
---|
220 |
|
---|
221 | sub read_file {
|
---|
222 | my $self = shift (@_);
|
---|
223 | my ($filename, $encoding, $language, $textref) = @_;
|
---|
224 |
|
---|
225 | my $metadata_table = $self->{'metadata_table'};
|
---|
226 |
|
---|
227 | my $rawtext = $metadata_table->{'rawtext'};
|
---|
228 |
|
---|
229 | delete $metadata_table->{'rawtext'};
|
---|
230 |
|
---|
231 | $$textref = $rawtext;
|
---|
232 | }
|
---|
233 |
|
---|
234 | sub read {
|
---|
235 | my $self = shift (@_);
|
---|
236 | my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
---|
237 |
|
---|
238 | my $outhandle = $self->{'outhandle'};
|
---|
239 |
|
---|
240 | return 0 if (defined $self->{'extra_blocks'}->{$file});
|
---|
241 |
|
---|
242 | # can we process this file??
|
---|
243 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
---|
244 | return undef unless $self->can_process_this_file($filename_full_path);
|
---|
245 |
|
---|
246 | $self->{'metadata_table'} = $metadata;
|
---|
247 |
|
---|
248 | my $lom_language = $metadata->{'lom_language'};
|
---|
249 |
|
---|
250 | my $store_input_encoding;
|
---|
251 | my $store_extract_language;
|
---|
252 | my $store_default_language;
|
---|
253 | my $store_default_encoding;
|
---|
254 |
|
---|
255 | if (defined $lom_language) {
|
---|
256 | delete $metadata->{'lom_language'};
|
---|
257 |
|
---|
258 | $store_input_encoding = $self->{'input_encoding'};
|
---|
259 | $store_extract_language = $self->{'extract_language'};
|
---|
260 | $store_default_language = $self->{'default_language'};
|
---|
261 | $store_default_encoding = $self->{'default_encoding'};
|
---|
262 |
|
---|
263 | $self->{'input_encoding'} = "utf8";
|
---|
264 | $self->{'extract_language'} = 0;
|
---|
265 | $self->{'default_language'} = $lom_language;
|
---|
266 | $self->{'default_encoding'} = "utf8";
|
---|
267 | }
|
---|
268 |
|
---|
269 | my $rv = $self->SUPER::read(@_);
|
---|
270 |
|
---|
271 | if (defined $lom_language) {
|
---|
272 | $self->{'input_encoding'} = $store_input_encoding;
|
---|
273 | $self->{'extract_language'} = $store_extract_language;
|
---|
274 | $self->{'default_language'} = $store_default_language;
|
---|
275 | $self->{'default_encoding'} = $store_default_encoding;
|
---|
276 | }
|
---|
277 |
|
---|
278 | $self->{'metadata_table'} = undef;
|
---|
279 |
|
---|
280 | return $rv;
|
---|
281 | }
|
---|
282 |
|
---|
283 | # do plugin specific processing of doc_obj
|
---|
284 | sub process {
|
---|
285 | my $self = shift (@_);
|
---|
286 | my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
|
---|
287 | my $outhandle = $self->{'outhandle'};
|
---|
288 |
|
---|
289 | my $cursection = $doc_obj->get_top_section();
|
---|
290 | $doc_obj->add_utf8_text($cursection, $$textref);
|
---|
291 |
|
---|
292 | return 1;
|
---|
293 | }
|
---|
294 |
|
---|
295 | sub Doctype {
|
---|
296 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
---|
297 |
|
---|
298 | my $root_tag = $self->{'root_tag'};
|
---|
299 |
|
---|
300 | if ($name !~ /$root_tag/) {
|
---|
301 | die "Root tag $name does not match regular expression $root_tag";
|
---|
302 | }
|
---|
303 | }
|
---|
304 |
|
---|
305 | sub StartTag {
|
---|
306 | my ($expat, $element) = @_;
|
---|
307 |
|
---|
308 | my %attr = %_;
|
---|
309 |
|
---|
310 | my $raw_tag = "<$element";
|
---|
311 | map { $raw_tag .= " $_=\"$attr{$_}\""; } keys %attr;
|
---|
312 | $raw_tag .= ">";
|
---|
313 |
|
---|
314 | if ($element =~ m/$self->{'root_tag'}/) {
|
---|
315 | $self->{'raw_text'} = $raw_tag;
|
---|
316 |
|
---|
317 | $self->{'saved_metadata'} = {};
|
---|
318 | $self->{'metaname_stack'} = [];
|
---|
319 | $self->{'lom_datatype'} = "";
|
---|
320 | $self->{'lom_language'} = undef;
|
---|
321 | $self->{'metadatatext'} = "<table class=\"metadata\" width=\"_pagewidth_\" >\n";
|
---|
322 | }
|
---|
323 | else {
|
---|
324 | my $xml_depth = scalar(@{$self->{'metaname_stack'}});
|
---|
325 | $self->{'raw_text'} .= "\n";
|
---|
326 | $self->{'raw_text'} .= " " x $xml_depth;
|
---|
327 | $self->{'raw_text'} .= $raw_tag;
|
---|
328 |
|
---|
329 | my $metaname_stack = $self->{'metaname_stack'};
|
---|
330 | push(@$metaname_stack,$element);
|
---|
331 | if (scalar(@$metaname_stack)==1) {
|
---|
332 | # top level LOM category
|
---|
333 | my $style = "class=\"metadata\"";
|
---|
334 | my $open_close
|
---|
335 | = "<a id=\"${element}opencloselink\" href=\"javascript:hideTBodyArea('$element')\">\n";
|
---|
336 | $open_close
|
---|
337 | .= "<img id=\"${element}openclose\" border=\"0\" src=\"_httpopenmdicon_\"></a>\n";
|
---|
338 |
|
---|
339 | my $header_line = " <tr $style ><th $style colspan=\"3\">$open_close \u$element</th></tr>\n";
|
---|
340 | my $md_tbody = "<tbody id=\"$element\">\n";
|
---|
341 |
|
---|
342 | $self->{'mdheader'} = $header_line;
|
---|
343 | $self->{'mdtbody'} = $md_tbody;
|
---|
344 | $self->{'mdtbody_text'} = "";
|
---|
345 | }
|
---|
346 | }
|
---|
347 | }
|
---|
348 |
|
---|
349 | sub EndTag {
|
---|
350 | my ($expat, $element) = @_;
|
---|
351 |
|
---|
352 | my $raw_tag = "</$element>";
|
---|
353 |
|
---|
354 | if ($element =~ m/$self->{'root_tag'}/) {
|
---|
355 | $self->{'raw_text'} .= $raw_tag;
|
---|
356 |
|
---|
357 | my $metadatatext = $self->{'metadatatext'};
|
---|
358 | $metadatatext .= "</table>";
|
---|
359 |
|
---|
360 | my $raw_text = $self->{'raw_text'};
|
---|
361 |
|
---|
362 | $self->{'saved_metadata'}->{'MetadataTable'} = $metadatatext;
|
---|
363 | $self->{'metadatatext'} = "";
|
---|
364 |
|
---|
365 | $self->{'saved_metadata'}->{'rawtext'} = $raw_text;
|
---|
366 | $self->{'raw_text'} = "";
|
---|
367 |
|
---|
368 | if (defined $self->{'lom_language'}) {
|
---|
369 | $self->{'saved_metadata'}->{'lom_language'} = $self->{'lom_language'};
|
---|
370 | $self->{'lom_language'} = undef;
|
---|
371 | }
|
---|
372 | }
|
---|
373 | else {
|
---|
374 | my $metaname_stack = $self->{'metaname_stack'};
|
---|
375 |
|
---|
376 | if (scalar(@$metaname_stack)==1) {
|
---|
377 | my $header_line = $self->{'mdheader'};
|
---|
378 | my $tbody_start = $self->{'mdtbody'};
|
---|
379 | my $tbody_text = $self->{'mdtbody_text'};
|
---|
380 | if ($tbody_text !~ m/^\s*$/s) {
|
---|
381 | my $tbody_end = "</tbody>\n";
|
---|
382 | my $table_chunk
|
---|
383 | = $header_line.$tbody_start.$tbody_text.$tbody_end;
|
---|
384 |
|
---|
385 | $self->{'metadatatext'} .= $table_chunk;
|
---|
386 | }
|
---|
387 | $self->{'mdtheader'} = "";
|
---|
388 | $self->{'mdtbody'} = "";
|
---|
389 | $self->{'mdtbody_text'} = "";
|
---|
390 | }
|
---|
391 |
|
---|
392 | pop(@$metaname_stack);
|
---|
393 |
|
---|
394 | my $xml_depth = scalar(@{$self->{'metaname_stack'}});
|
---|
395 | $self->{'raw_text'} .= "\n";
|
---|
396 | $self->{'raw_text'} .= " " x $xml_depth;
|
---|
397 | $self->{'raw_text'} .= $raw_tag;
|
---|
398 | }
|
---|
399 | }
|
---|
400 |
|
---|
401 | sub process_datatype_info
|
---|
402 | {
|
---|
403 | my $self = shift(@_);
|
---|
404 | my ($metaname_stack,$md_content) = @_;
|
---|
405 |
|
---|
406 | my @without_dt_stack = @$metaname_stack; # without datatype stack
|
---|
407 |
|
---|
408 | my $innermost_element = $without_dt_stack[$#without_dt_stack];
|
---|
409 |
|
---|
410 | # Loose last item if encoding datatype information
|
---|
411 | if ($innermost_element =~ m/^(lang)?string$/) {
|
---|
412 | $self->{'lom_datatype'} = $innermost_element;
|
---|
413 |
|
---|
414 | pop @without_dt_stack;
|
---|
415 | $innermost_element = $without_dt_stack[$#without_dt_stack];
|
---|
416 | }
|
---|
417 | elsif ($innermost_element =~ m/^date(Time)?$/i) {
|
---|
418 | if ($innermost_element =~ m/^date$/i) {
|
---|
419 | $self->{'lom_datatype'} = "dateTime";
|
---|
420 | }
|
---|
421 | else {
|
---|
422 | $self->{'lom_datatype'} = $innermost_element;
|
---|
423 |
|
---|
424 | pop @without_dt_stack;
|
---|
425 | $innermost_element = $without_dt_stack[$#without_dt_stack];
|
---|
426 | }
|
---|
427 |
|
---|
428 | if ($md_content =~ m/^(\d{1,2})\s*(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)\w*\s*(\d{4})/i) {
|
---|
429 | my ($day,$mon,$year) = ($1,$2,$3);
|
---|
430 |
|
---|
431 | my %month_lookup = ( 'jan' => 1, 'feb' => 2, 'mar' => 3,
|
---|
432 | 'apr' => 4, 'may' => 5, 'jun' => 6,
|
---|
433 | 'jul' => 7, 'aug' => 8, 'sep' => 9,
|
---|
434 | 'oct' => 10, 'nov' => 11, 'dec' => 12 );
|
---|
435 |
|
---|
436 | my $mon_num = $month_lookup{lc($mon)};
|
---|
437 |
|
---|
438 | $md_content = sprintf("%d%02d%02d",$year,$mon_num,$day);
|
---|
439 | }
|
---|
440 |
|
---|
441 | $md_content =~ s/\-//g;
|
---|
442 | }
|
---|
443 |
|
---|
444 | if ($innermost_element eq "source") {
|
---|
445 | $self->{'lom_source'} = $md_content;
|
---|
446 | }
|
---|
447 | elsif ($innermost_element eq "value") {
|
---|
448 | $self->{'lom_value'} = $md_content;
|
---|
449 | }
|
---|
450 |
|
---|
451 | return (\@without_dt_stack,$innermost_element,$md_content);
|
---|
452 | }
|
---|
453 |
|
---|
454 | sub reset_datatype_info
|
---|
455 | {
|
---|
456 | my $self = shift(@_);
|
---|
457 |
|
---|
458 | $self->{'lom_datatype'} = "";
|
---|
459 | }
|
---|
460 |
|
---|
461 |
|
---|
462 | sub pretty_print_text
|
---|
463 | {
|
---|
464 | my $self = shift(@_);
|
---|
465 |
|
---|
466 | my ($pretty_print_text) = @_;
|
---|
467 |
|
---|
468 | ## $metavalue_utf8 = &util::hyperlink_text($metavalue_utf8);
|
---|
469 | $pretty_print_text = &util::hyperlink_text($pretty_print_text);
|
---|
470 |
|
---|
471 | #### $pretty_print_text =~ s/(BEGIN:vCard.*END:vCard)/<pre>$1<\/pre>/sg;
|
---|
472 |
|
---|
473 | if ($self->{'lom_datatype'} eq "dateTime") {
|
---|
474 | if ($pretty_print_text =~ m/^(\d{4})(\d{2})(\d{2})$/) {
|
---|
475 | $pretty_print_text = "$1-$2-$3";
|
---|
476 | }
|
---|
477 | }
|
---|
478 |
|
---|
479 | return $pretty_print_text;
|
---|
480 | }
|
---|
481 |
|
---|
482 | sub pretty_print_table_tr
|
---|
483 | {
|
---|
484 | my $self = shift (@_);
|
---|
485 | my ($without_dt_stack) = @_;
|
---|
486 |
|
---|
487 | my $style = "class=\"metadata\"";
|
---|
488 |
|
---|
489 | my $innermost_element = $without_dt_stack->[scalar(@$without_dt_stack)-1];
|
---|
490 | my $outermost_element = $without_dt_stack->[0];
|
---|
491 |
|
---|
492 | # Loose top level stack item (already named in pretty print table)
|
---|
493 | my @pretty_print_stack = @$without_dt_stack;
|
---|
494 | shift @pretty_print_stack;
|
---|
495 |
|
---|
496 | if ($innermost_element eq "source") {
|
---|
497 | return if (!defined $self->{'lom_value'});
|
---|
498 | }
|
---|
499 |
|
---|
500 | if ($innermost_element eq "value") {
|
---|
501 | return if (!defined $self->{'lom_source'});
|
---|
502 | }
|
---|
503 |
|
---|
504 | my $pretty_print_text = "";
|
---|
505 |
|
---|
506 | if (($innermost_element eq "value") || ($innermost_element eq "source")) {
|
---|
507 | my $source = $self->{'lom_source'};
|
---|
508 | my $value = $self->pretty_print_text($self->{'lom_value'});
|
---|
509 |
|
---|
510 | $self->{'lom_source'} = undef;
|
---|
511 | $self->{'lom_value'} = undef;
|
---|
512 |
|
---|
513 | pop @pretty_print_stack;
|
---|
514 |
|
---|
515 | $pretty_print_text = "<td $style>$source</td><td $style>$value</td>";
|
---|
516 | }
|
---|
517 | else {
|
---|
518 | $pretty_print_text = $self->pretty_print_text($_);
|
---|
519 | $pretty_print_text = "<td $style colspan=2>$pretty_print_text</td>";
|
---|
520 | }
|
---|
521 | my $pretty_print_fmn = join(' : ',map { "\u$_"; } @pretty_print_stack);
|
---|
522 |
|
---|
523 |
|
---|
524 | # my $tr_attr = "id=\"$outermost_element\" style=\"display:block;\"";
|
---|
525 | my $tr_attr = "$style id=\"$outermost_element\"";
|
---|
526 |
|
---|
527 | my $mdtext_line = " <tr $tr_attr><td $style><nobr>$pretty_print_fmn</nobr></td>$pretty_print_text</tr>\n";
|
---|
528 | $self->{'mdtbody_text'} .= $mdtext_line;
|
---|
529 | }
|
---|
530 |
|
---|
531 |
|
---|
532 | sub check_for_language
|
---|
533 | {
|
---|
534 | my $self = shift(@_);
|
---|
535 | my ($innermost_element,$md_content) = @_;
|
---|
536 |
|
---|
537 | # Look for 'language' tag
|
---|
538 | if ($innermost_element eq "language") {
|
---|
539 | my $lom_lang = $self->{'lom_language'};
|
---|
540 |
|
---|
541 | if (defined $lom_lang) {
|
---|
542 | my $new_lom_lang = $md_content;
|
---|
543 | $new_lom_lang =~ s/-.*//; # remove endings like -US or -GB
|
---|
544 |
|
---|
545 | if ($lom_lang ne $new_lom_lang) {
|
---|
546 | my $outhandle = $self->{'outhandle'};
|
---|
547 |
|
---|
548 | print $outhandle "Warning: Conflicting general language in record\n";
|
---|
549 | print $outhandle " $new_lom_lang (previous value for language = $lom_lang)\n";
|
---|
550 | }
|
---|
551 | # otherwise, existing value OK => do nothing
|
---|
552 | }
|
---|
553 | else {
|
---|
554 | $lom_lang = $md_content;
|
---|
555 | $lom_lang =~ s/-.*//; # remove endings like -US or -GB
|
---|
556 |
|
---|
557 | $self->{'lom_language'} = $lom_lang;
|
---|
558 | }
|
---|
559 | }
|
---|
560 | }
|
---|
561 |
|
---|
562 | sub found_specific_identifier
|
---|
563 | {
|
---|
564 | my $self = shift(@_);
|
---|
565 | my ($specific_id,$full_mname,$md_content) = @_;
|
---|
566 |
|
---|
567 | my $found_id = 0;
|
---|
568 | if ($full_mname eq $specific_id) {
|
---|
569 | if ($md_content =~ m/^(http|ftp):/) {
|
---|
570 | $found_id = 1;
|
---|
571 | }
|
---|
572 | }
|
---|
573 |
|
---|
574 | return $found_id;
|
---|
575 | }
|
---|
576 |
|
---|
577 | sub download_srcdoc
|
---|
578 | {
|
---|
579 | my $self = shift(@_);
|
---|
580 | my ($doc_url) = @_;
|
---|
581 |
|
---|
582 | my $outhandle = $self->{'outhandle'};
|
---|
583 | my $output_dir = $self->{'output_dir'};
|
---|
584 |
|
---|
585 | $output_dir = &FileUtils::filenameConcatenate($output_dir,"_gsdldown.all");
|
---|
586 |
|
---|
587 | if (! -d $output_dir) {
|
---|
588 | mkdir $output_dir;
|
---|
589 | }
|
---|
590 |
|
---|
591 | my $re_dirsep = &util::get_re_dirsep();
|
---|
592 | my $os_dirsep = &util::get_dirsep();
|
---|
593 |
|
---|
594 | my $file_url = $doc_url;
|
---|
595 | $file_url =~ s/$re_dirsep/$os_dirsep/g;
|
---|
596 | $file_url =~ s/^(http|ftp):\/\///;
|
---|
597 | $file_url .= "index.html" if ($file_url =~ m/\/$/);
|
---|
598 |
|
---|
599 | my $full_file_url = &FileUtils::filenameConcatenate($output_dir,$file_url);
|
---|
600 | # the path to srcdoc will be used later in extrametadata to associate
|
---|
601 | # the lom metadata with the document. Needs to be relative to current
|
---|
602 | # directory.
|
---|
603 | my $srcdoc_path = &FileUtils::filenameConcatenate("_gsdldown.all", $file_url);
|
---|
604 | my $check_timestamp = $self->{'check_timestamp'};
|
---|
605 | my $status;
|
---|
606 |
|
---|
607 | if (($check_timestamp) || (!$check_timestamp && !-e $full_file_url)) {
|
---|
608 | if (!-e $full_file_url) {
|
---|
609 | print $outhandle "Mirroring $doc_url\n";
|
---|
610 | }
|
---|
611 | else {
|
---|
612 | print $outhandle "Checking to see if update needed for $doc_url\n";
|
---|
613 | }
|
---|
614 |
|
---|
615 | # on linux, if we pass an absolute path as -P arg to wget, then it
|
---|
616 | # stuffs up the
|
---|
617 | # URL rewriting in the file. Need a relative path or none, so now
|
---|
618 | # we change working directory first.
|
---|
619 | my $changed_dir = 0;
|
---|
620 | my $current_dir = cwd();
|
---|
621 | my $wget_cmd = "";
|
---|
622 | if ($ENV{'GSDLOS'} ne "windows") {
|
---|
623 | $changed_dir = 1;
|
---|
624 |
|
---|
625 | chdir "$output_dir";
|
---|
626 | $wget_cmd = "wget -nv --timestamping -k -p \"$doc_url\"";
|
---|
627 | } else {
|
---|
628 | $wget_cmd = "wget -nv -P \"$output_dir\" --timestamping -k -p \"$doc_url\"";
|
---|
629 | }
|
---|
630 | ##print STDERR "**** wget = $wget_cmd\n";
|
---|
631 |
|
---|
632 | # the wget binary is dependent on the gnomelib_env (particularly lib/libiconv2.dylib) being set, particularly on Mac Lions (android too?)
|
---|
633 | &util::set_gnomelib_env(); # this will set the gnomelib env once for each subshell launched, by first checking if GEXTGNOME is not already set
|
---|
634 |
|
---|
635 | $status = system($wget_cmd);
|
---|
636 | if ($changed_dir) {
|
---|
637 | chdir $current_dir;
|
---|
638 | }
|
---|
639 | if ($status==0) {
|
---|
640 | $self->{'lom_srcdoc'} = $srcdoc_path;
|
---|
641 | }
|
---|
642 | else {
|
---|
643 | $self->{'lom_srcdoc'} = undef;
|
---|
644 | print $outhandle "Error: failed to execute $wget_cmd\n";
|
---|
645 | }
|
---|
646 | }
|
---|
647 | else {
|
---|
648 | # not time-stamping and file already exists
|
---|
649 | $status=0;
|
---|
650 | $self->{'lom_srcdoc'} = $srcdoc_path;
|
---|
651 | }
|
---|
652 |
|
---|
653 | return $status==0;
|
---|
654 |
|
---|
655 | }
|
---|
656 |
|
---|
657 |
|
---|
658 | sub check_for_identifier
|
---|
659 | {
|
---|
660 | my $self = shift(@_);
|
---|
661 | my ($full_mname,$md_content) = @_;
|
---|
662 |
|
---|
663 | my $success = 0;
|
---|
664 |
|
---|
665 | my $download_re = $self->{'download_srcdocs'};
|
---|
666 | if (($download_re ne "") && $md_content =~ m/$download_re/) {
|
---|
667 |
|
---|
668 | if ($self->found_specific_identifier("general^identifier^entry",$full_mname,$md_content)) {
|
---|
669 | $success = $self->download_srcdoc($md_content);
|
---|
670 | }
|
---|
671 |
|
---|
672 | if (!$success) {
|
---|
673 | if ($self->found_specific_identifier("technical^location",$full_mname,$md_content)) {
|
---|
674 | $success = $self->download_srcdoc($md_content);
|
---|
675 | }
|
---|
676 | }
|
---|
677 | }
|
---|
678 |
|
---|
679 | return $success;
|
---|
680 | }
|
---|
681 |
|
---|
682 |
|
---|
683 | sub Text {
|
---|
684 | if ($_ !~ m/^\s*$/) {
|
---|
685 | #
|
---|
686 | # Work out indentations and line wraps for raw XML
|
---|
687 | #
|
---|
688 | my $xml_depth = scalar(@{$self->{'metaname_stack'}})+1;
|
---|
689 | my $indent = " " x $xml_depth;
|
---|
690 |
|
---|
691 | my $formatted_text = "\n".$_;
|
---|
692 |
|
---|
693 | # break into lines < 80 chars on space
|
---|
694 | $formatted_text =~ s/(.{50,80})\s+/$1\n/mg;
|
---|
695 | $formatted_text =~ s/^/$indent/mg;
|
---|
696 | ## $formatted_text =~ s/\s+$//s;
|
---|
697 |
|
---|
698 | $self->{'raw_text'} .= $formatted_text;
|
---|
699 | }
|
---|
700 |
|
---|
701 | my $metaname_stack = $self->{'metaname_stack'};
|
---|
702 | if (($_ !~ /^\s*$/) && (scalar(@$metaname_stack)>0)) {
|
---|
703 |
|
---|
704 | my ($without_dt_stack,$innermost_element,$md_content)
|
---|
705 | = $self->process_datatype_info($metaname_stack,$_);
|
---|
706 |
|
---|
707 | $self->pretty_print_table_tr($without_dt_stack);
|
---|
708 |
|
---|
709 | my $full_mname = join('^',@{$without_dt_stack});
|
---|
710 | $self->set_filere_metadata(lc($full_mname),$md_content);
|
---|
711 |
|
---|
712 | $self->check_for_language($innermost_element,$md_content);
|
---|
713 | $self->check_for_identifier($full_mname,$md_content); # source doc
|
---|
714 |
|
---|
715 | $self->reset_datatype_info();
|
---|
716 | }
|
---|
717 | }
|
---|
718 |
|
---|
719 | # This Char function overrides the one in XML::Parser::Stream to overcome a
|
---|
720 | # problem where $expat->{Text} is treated as the return value, slowing
|
---|
721 | # things down significantly in some cases.
|
---|
722 | sub Char {
|
---|
723 | $_[0]->{'Text'} .= $_[1];
|
---|
724 | return undef;
|
---|
725 | }
|
---|
726 |
|
---|
727 | 1;
|
---|